Interdependence of Tissue Temperature, Cavitation, and Displacement Imaging During Focused Ultrasound Nerve Sonication

Abstract

Focused ultrasound (FUS) peripheral neuromodulation has been linked to nerve displacement caused by the acoustic radiation force; however, the roles of cavitation and temperature accumulation on nerve modulation are less clear, as are the relationships between these three mechanisms of action. Temperature directly changes tissue stiffness and viscosity. Viscoelastic properties have been shown to affect cavitation thresholds in both theoretical and ex vivo models, but the direct effect of temperature on cavitation has not been investigated in vivo. Here, cavitation and tissue displacement were simultaneously mapped in response to baseline tissue temperatures of either 30°C or 38°C during sciatic nerve sonication in mice. In each mouse, the sciatic nerve was repeatedly sonicated at 1.1 MHz, 4 MPa peak-negative pressure, 5 ms pulse duration, and either 15 Hz or 30 Hz pulse repetition frequency for 10 s at each tissue temperature. Cavitation increased by 1.8-4.5 dB at a tissue temperature of 38°C compared to 30°C, as measured both by passive cavitation images and cavitation doses. Tissue displacement also increased by 1.3-1.8 μm at baseline temperatures of 38°C compared to 30°C. Histological findings indicated small increases in red blood cell extravasation in the 38°C baseline temperature condition compared to 30°C at both PRFs. A strong positive correlation was found between the inertial cavitation dose and displacement imaging noise, indicating the potential ability of displacement imaging to simultaneously detect inertial cavitation in vivo. Overall, tissue temperature was found to modulate both in vivo cavitation and tissue displacement, and thus both tissue temperature and cavitation can be monitored during FUS to ensure both safety and efficiency.

I. Introduction

FOCUSED ultrasound (FUS) has been shown to be capable of inducing excitatory and inhibitory effects in both in vivo and ex vivo models of the central [1]-[11] and peripheral [12]-[22] nervous systems. This method of stimulation, known as FUS neuromodulation, provides a method for targeting nerves both non-invasively and with high spatial precision. Ultrasound can produce thermal or mechanical effects on tissues, or a combination of the two, depending on the parameters employed. Two mechanical phenomena resulting from FUS are the acoustic radiation force and cavitation. Neural excitation has been linked directly to the acoustic radiation force [23], or indirectly to tissue displacement caused by nonlinear acoustic radiation force interaction with tissue [19], but the potential role of cavitation as a neuromodulation mechanism in vivo is unclear, as is the relationship between cavitation and displacement.

Cavitation denotes the generation and/or subsequent growth of gas-filled nuclei within a medium—tissue, for example—during the rarefaction phase of the acoustic cycle [24], [25]. The acoustic pressure amplitude greatly affects the onset of cavitation, with higher peak negative pressures increasing the probability of cavitation. Direct peripheral excitation relies on the use of high sonication pressures, typically at least 1 MPa [26] but potentially as high as 28 MPa [19]. The use of these high pressures emphasizes cavitation both as a potential mechanism for peripheral neuromodulation as well as a safety consideration [26]. Previous studies implicating cavitation as a mechanism have focused on modeling [27]-[29] or ex vivo experimentation [14], [16] whereas the role of cavitation as an in vivo neuromodulation mechanism is largely unexplored experimentally.

One potential factor influencing the likelihood of cavitation in vivo is tissue temperature. Core body temperature naturally varies between subjects as much as 2.2°C [30]. Heat can also accumulate in the tissue during FUS therapies, resulting in changing tissue temperatures over time. Previous studies have illustrated the influence of tissue shear modulus, a measurement of the rigidity or softness of a tissue, and viscosity on the occurrence of inertial cavitation [31], [32]. In a modeling study, increasing the rigidity of tissue increased the threshold of inertial cavitation, assuming preexisting spherical bubbles in the elastic medium [31]. Similarly, an in vitro histotripsy— an ablative method of fractionating tissue by controlling acoustic cavitation— study found impeded bubble expansion from cavitation in stiffer tissues [32], illustrating the influence of tissue stiffness on cavitation dynamics. Tissue temperature changes the stiffness of tissue; specifically, heating tissue from about 25°C to 57°C softens tissue and lowers the shear modulus [33]. Thus, temperature may alter cavitation activity as tissue viscoelastic properties shift in response to heating and cooling. There is already support for this phenomenon in ex vivo and in vitro studies. Increasing the temperature of water has been shown to lower the intrinsic cavitation threshold [34] and heating ex vivo tissues has been shown to produce greater lesions in response to histotripsy when compared to stiffer, non-heated tissues [35]. The effect of increasing tissue temperature on cavitation activity in vivo, however, has not been well characterized.

Just as cavitation thresholds could be altered in response to shifting viscoelastic properties during tissue heating, temperature modulation could also affect the resulting nerve displacement per FUS pulse. In FUS neuromodulation, where displacement imaging capable of real-time monitoring has been used to correlate nerve displacement with both motor activation [19] and pain suppression [21] and where cavitation may be an additional activation mechanism, tissue temperature modulation could dictate the success of treatment. Monitoring both displacement and cavitation during the same FUS procedure may also better elucidate the role of these mechanisms during neuromodulation.

Thus, in the present study, displacement imaging was interleaved with passive cavitation acquisition to estimate both mechanisms during a single FUS procedure in the in vivo mouse hindlimb. Specifically, cavitation and displacement were both acquired during nerve sonication and the effect of temperature on each phenomenon was investigated. It was predicted that increased tissue temperatures would result in greater cavitation activity and nerve displacement compared to lower tissue temperatures. The sciatic nerve was targeted with FUS and cavitation and displacements were recorded during each FUS procedure at temperatures of 30°C and 38°C.

II. Methods

A. Animal Preparation

All animal procedures were approved by the Institutional Animal Care and Use Committee of Columbia University. Twenty-seven (n = 27; n = 8 male and n = 19 female) wild type C57BL/6J mice weighing between 17.9 g and 25.0 g were used in all experiments. Mice were anesthetized with isoflurane: 2% during preparation and 1.5-2% during the procedure. Both hindlimbs on each mouse were shaved and dehaired using a depilatory cream. A heating pad was used on the upper half of the mouse body to maintain proper core body temperature (36-37°C) throughout the experiment. Anesthesia depth was monitored every 15 minutes by observing respiration and pedal reflexes.

B. Experimental Setup

The sciatic nerve in mice was sonicated at a controlled baseline temperature using the experimental setup shown in Fig. 1 (a). A 1.1 MHz, 4-element annular array (Sonic Concepts Inc., USA) applied pulsed ultrasound to the mouse hindlimb. This annular array had a focal depth of 51.74 mm and an active diameter of 64 mm. B-mode imaging, passive cavitation acquisition, and displacement imaging were conducted at a center frequency of 7.8 MHz with a concentric P12-5 transducer (104 element phased array; ATL Philips, Netherlands) placed through the 28.7 mm opening of the FUS transducer. The lower half of each mouse was placed in a temperature-controlled water bath to modulate the hindlimb nerve and muscle tissue temperature during sonication.

Fig. 1.

(a) Experimental set-up demonstrating the coaxial placement of the 1.1 MHz FUS transducer and the 7.8 MHz imaging transducer. The sciatic nerve of the mouse was targeted during each sonication. The hindlimb of the mouse was placed in a water bath to control the tissue temperature. (b) Hydrophone measurement of the axial and lateral beam profile of the 1.1 MHz FUS transducer, illustrating the beam size in both axial and lateral directions. (c) Example displacement targeting of the sciatic nerve. The sciatic nerve is marked in each image with a dotted white line. The push and relaxation from FUS sonication occurs on the sciatic nerve, indicating successful targeting.

C. Acoustic Parameters

The sciatic nerve of each mouse was sonicated at a pressure of 4 MPa peak-negative pressure (PNP). Pulse durations of 5 ms were repeated at a pulse repetition frequency (PRF) of either 15 Hz or 30 Hz for a total sonication duration of 10 s. These acoustic parameters have been used by our group in human peripheral neuromodulation studies for pain suppression and thus were similarly selected here. Lateral and axial beam profiles demonstrated full-width half-max (FWHM) beam lengths of 1.7 mm laterally and 17 mm axially (Fig. 1) (c).

D. Experimental Design

In this study design, 11 consecutive sonications were performed at a constant baseline temperature of either 30°C or 38°C in the mouse hindlimb to investigate changes in in vivo tissue cavitation and displacement with temperature modulation. The lowest baseline temperature of 30°C was chosen as it was the temperature found in the mouse leg in previous experiments when acoustic coupling gel was placed on the leg, whereas the highest temperature of 38°C was chosen as it was within the normal mouse body temperature (36.5-38°C). Cavitation activity and tissue displacement were recorded during each sonication. For each mouse, one leg received sonications at 30°C and the second leg received sonications at 38°C to control for between-subject differences. In each mouse, all sonications occurred in one leg at 30°C and then all sonications occurred in the second leg at 38°C, although each mouse was randomly assigned the right or left as the starting leg. In each mouse, sonications were performed either at 30 Hz PRF (n = 7 mice) or 15 Hz PRF (n = 7 mice). At least two minutes were allotted between each 10 s sonication for any accumulated temperature to dissipate and for the hindlimb to stabilize back to the baseline temperature.

A second, control experiment was also performed to investigate potential differences in in vivo tissue cavitation arising from hindlimb sonication order. In these control mice, both hindlimbs were kept at a baseline temperature of 30°C. For each mouse, one leg received all sonications at 30°C and then the second leg received all sonications also at 30°C, although in each mouse it was randomly determined whether the right or left leg would be the starting leg. Cavitation was then separately analyzed in the first and second legs sonicated. As with the first study design, two PRFs of 30 Hz (n = 4 mice) and 15 Hz (n = 4 mice) were investigated. All other procedures remained the same between the two study designs.

E. Tissue Temperature Regulation

To control for tissue temperature of the mouse hindlimb, the lower half of the mouse body was placed in a water bath. The heat rejecting (“hot”) side of a Peltier device (TEC1-12706; Lexiesxue, USA) was coupled to the water bath and the water temperature was modulated by controlling the current delivered through the Peltier device. A thermocouple inserted in high proximity to the FUS focal spot recorded the temperature in the mouse hindlimb before and during sonication (Fig. 2) (a). To achieve the desired baseline temperature in the hindlimb prior to sonication, the water bath was heated or cooled until the desired temperature was measured by the thermocouple in the hindlimb.

Fig. 2.

(a) Location of thermocouple outside of the FUS focus during the experimental procedure. The orange dot represents the location of maximum heating at the center of the FUS focus. (b) Illustrative temperature increases at the FUS focus and at the thermocouple 3 mm from the FUS focus at a PRF of 15 Hz. (c) Average temperature increases at the FUS focus at 15 Hz and 30 Hz PRF, as calculated by adding the experimentally derived difference between the temperature increase at and outside of the FUS focus.

The thermocouple was placed outside the focus during each experiment to ensure the tip did not interfere with cavitation recording. To calculate the expected temperature increase at the FUS focus from sonication, mice (n = 2) underwent 10 s sonications where the temperature at and outside of the focus was measured with two thermocouples (Fig. 2) (c). The average temperature difference between these two thermocouples was measured for distances 1.5-5 mm away from the focus. During the two experimental studies, the temperature at the focus was estimated by adding the expected temperature increase according to the distance of the thermocouple from the focus.

F. Displacement Imaging

Displacement imaging following the procedure developed by Lee et al. [19] was used to confirm successful sciatic nerve targeting prior to sonication (Fig. 1) (b). Targeting was conducted at a pressure of 3 MPa PNP using a single, 1 ms pulse to avoid inducing damage or inertial cavitation.

Either displacement imaging or passive cavitation recording occurred simultaneously with each 5 ms FUS pulse throughout the 10 s sonication period. Displacement estimation was conducted during the third, middle, and final FUS pulses, whereas passive cavitation recording occurred throughout all other FUS pulses during each 10 s sonication (Fig. 3) (a). Plane wave transmit and receive imaging at a rate of 4 kHz for displacement estimation occurred for 0.5 ms before, 5 ms during, and 5 ms after the FUS pulse (Fig. 3) (b).

Fig. 3.

(a) Illustration of the FUS sequence implemented during the 10 s sonication. Each blue and orange bar represent a 5 ms FUS pulse. Blue bars represent cavitation acquisition during the 5 ms pulse, whereas orange bars represent displacement imaging. Three displacement acquisitions occur throughout the sonication duration, while cavitation recording occurs during all other pulses. The dark lines within each orange or blue block represent the imaging sequence occurring within each 5 ms pulse. Receive-only cavitation recording occurs at a rate of 1 kHz, whereas transmit and receive pulses for displacement imaging occur at a rate of 4 kHz. (b) An illustrative displacement imaging trace averaged within an ROI around the sciatic nerve, showing both the push during the FUS-on time and the subsequent relaxation after the FUS pulse. (c) Flowchart illustrating the data processing steps taken during cavitation post-processing.

Displacement traces were computed for each 10 s sonication by taking the average interframe displacement per frame within a specified region of interest (ROI) from −0.85 to 0.85 mm laterally and 28 to 32 mm axially. This ROI was chosen based on the lateral width of the FUS transducer focus (Fig. 1) (c) and the region of tissue displaced axially as seen in the displacement images (Fig. 1) (b). The estimated displacement for each sonication was then taken as the average of the interframe displacement within the 5 ms FUS pulse (orange portion of Fig. 3 (b)). Displacement estimations from both the temperature-PRF and sonication order-PRF studies were included in the displacement versus temperature analysis. Four hundred and twenty-nine (n = 429) displacement estimations at 30°C were analyzed for 15 Hz and 30 Hz PRF and one hundred and sixty-five (n = 165) displacement estimations at 38°C were analyzed for 15 Hz and 30 Hz PRF.

The coefficient of variation (CV) was calculated for each displacement trace during the FUS on-time as a measure of noise. The equation used to calculate the CV was: CV=σ/μ [36], where σ represents the mean interframe displacement during the 5 ms FUS on-time and μ represents the standard deviation during the same time period.

G. Passive Cavitation Imaging

During each 10 s sonication, cavitation was passively acquired (receive-only) by the concentrically aligned P12-5 array (Fig. 1) (a). For each 5 ms pulse in the 10 s pulse train, 5 frames of cavitation data were acquired during the pulse and 1 baseline frame of cavitation data was acquired after the pulse at a rate of 1 kHz. The cavitation signals were post-processed with a frequency-domain time exposure acoustics (TEA) algorithm [37] and a frequency-based passive cavitation image (PCI) was calculated for each frame of data. An averaged PCI was also calculated for each total sonication. Each averaged PCI was overlaid onto a representative B-mode image from the respective parameter set. Stable cavitation was separated by taking a 200 kHz window around each 1.1 MHz harmonic from 2.2 MHz to 12.1 MHz. This window corresponded to a −6 dB width around each harmonic peak [38]. Inertial cavitation was separated by taking the range of frequencies after exclusion of the harmonics [38]. This resulted in a 900 kHz window between each windowed 1.1 MHz harmonic; for example, from 2.3 MHz to 3.2 MHz between the second and third harmonics.

The time-averaged inertial cavitation dose (ICD) and stable cavitation dose (SCD) were calculated by first taking the root mean square (RMS) across frequency of the separated harmonic and broadband data. This was performed for the baseline (FUS-off) and pulse (FUS-on) frames separately. Next, for each frame of this broadband and harmonic data, a spatial sum of the cavitation RMS amplitude was taken over a region of interest (ROI) from −0.85 to 0.85 mm laterally and 28 to 32 mm axially, as described in section 2.F. The spatially summed data in each frame was then converted to decibels. A single average baseline ICD and SCD were calculated from the FUS-off frames and subsequently subtracted from the FUS-on ICD and SCD values. Lastly, the processed stable and inertial baseline-subtracted doses for each sonication were averaged across frames to create the time-averaged cavitation doses. Average PCIs of each experimental group of mice were computed to compare the PCI across temperature-PRF or sonication order-PRF parameter combinations. This data processing workflow is depicted in Fig. 3 (c).

H. Histological Analysis

The safety of these high pressure FUS parameters at both PRFs and baseline tissue temperatures were analyzed in three mice (n = 3). Following the sonication procedure previously described, one mouse received FUS at 15 Hz PRF and the second mouse received FUS at 30 Hz PRF. In both mice, the 30°C baseline tissue temperature condition occurred in one hindlimb first, followed by the 38°C baseline tissue temperature condition in the other hindlimb. The third animal received sham sonication and a positive control condition, in separate hindlimbs, with the positive control consisting of FUS at 7.1 MPa, 15 ms PD, 50 Hz PRF, and a 2 min sonication duration. After sonication, the hindlimbs were immediately excised and fixed with 4% Paraformaldehyde (PFA) for 24 hours. Subsequently, the hindlimbs were placed in 70% ethanol (EtOH) for 24 hours before sectioning and Hematoxylin and Eosin (H&E) staining. Red blood cell extravasation, inflammation and swelling, and muscle tissue abnormalities were used as indication of damage.

I. Statistical Analysis

Statistical testing was performed in Prism 6 (GraphPad, San Diego, CA). Parametric unpaired two-tailed t-tests were conducted for the following data sets at both 15 and 30 Hz PRF: ICD and SCD across 30°C and 38°C baseline tissue temperature, ICD and SCD across hindlimb sonication order, PCI cavitation values across 30°C and 38°C baseline tissue temperature, PCI cavitation values across hindlimb sonication order, FUS-induced tissue temperature increases, and estimated tissue displacement across 30°C and 38°C baseline tissue temperature.

Pearson correlation coefficient calculation and exponential modeling were conducted in MATLAB (MathWorks, Natick, MA) to compare ICD and displacement CV values across all sonications. Coefficients for an exponential model were calculated such that they maximized the R² value of the following equation when applied to the data:

$$\text{CV}=\text{A}\cdot \text{exp}(\text{B}\cdot \text{ICD}[\text{dB}]-\text{C})+\text{D}$$ (1)

III. Results

A. Leg Tissue Temperature Increases Due to Focused Ultrasound

Along with the baseline tissue temperature dictated by the temperature of the water bath, the sciatic nerve and surrounding muscle and fat also underwent temperature changes from the FUS sonication. As depicted in Fig. 2 (c), these tissues experienced a temperature increase of 1.7 ± 0.8 °C as a result of 10 s sonication at 15 Hz PRF (n = 230 sonications) and 3.0 ± 0.6 °C at 30 Hz PRF (n = 235 sonications) (p < .0001, unpaired two-tailed t-test). At both PRFs, the temperature rise at a baseline temperature of 38°C was slightly lower than the rise found at 30°C. At 15 Hz PRF, a temperature increase of 1.7 ± 0.8 °C occurred at 30°C compared to a temperature increase of 1.6 ± 0.6°C at 38°C (p = 0.4, unpaired two-tailed t-test). At 30 Hz PRF, a temperature increase of 3.0 ± 0.6 °C occurred at 30°C compared to a temperature increase of 2.9 ± 0.6°C at 38°C (p = 0.6, unpaired two-tailed t-test). In both cases, however, this difference was not statistically significant.

B. Increasing Baseline Tissue Temperatures Increases Average Cavitation Activity

To compare changes in cavitation activity across both baseline tissue temperature and PRF, a total average PCI across seventy-seven (n = 77) sonications per average PCI was calculated for each tissue temperature-PRF pair. Fig. 4 (a) and (b) show these averaged PCI for 15 Hz and 30 Hz PRF at baseline tissue temperatures of 30°C and 38°C. For both PRFs, the average cavitation activity increased as the baseline tissue temperature went from 30°C to 38°C, as indicated by the increased intensity in the 38°C PCI compared to the 30°C PCI. To quantify this increase in PCI across temperatures, the average cavitation intensity for each individual PCI was calculated within the ROI shown. At 15 Hz PRF, the cavitation intensity within the ROI increased by 3.1 dB from 30°C to 38°C (56.9 ± 7.4 dB at 30°C vs. 60.0 ± 9.1 dB at 38°C; p < 0.05). A similar increase of 3.4 dB cavitation intensity within the ROI occurred between 30°C and 38°C at 30 Hz PRF (56.2 ± 7.1 dB at 30°C vs. 59.7 ± 3.9 dB at 38°C; p < 0.0001). No significant differences in cavitation intensity were found at 30°C between the 15 Hz and 30 Hz PRF PCI, nor at 38°C (p > 0.05).

Fig. 4.

Averaged PCI for each baseline tissue temperature at (a) 15 Hz PRF and (b) 30 Hz PRF. The location of the hindlimb and the sciatic nerve are indicated in each PCI. The dotted black box represents the ROI from −0.85 mm to 0.85 mm laterally and 28 to 32 mm axially. PCI intensity increases from 30°C to 38°C for both 15 Hz and 30 Hz PRF. The axial intensity profiles illustrate this PCI intensity at 30 mm axially for each PCI at (c) 15 Hz PRF and (d) 30 Hz PRF with the standard deviation shown.

Lateral intensity profiles obtained at the FUS focus of 30 mm axially demonstrate changes in the physical shape of the PCI between temperature and PRF parameters (Fig. 4) (c) and (d). In each intensity profile, the change between the 30°C and 38°C curves was determined by taking the FWHM for each curve and then calculating the percent change between the FWHM values. At 15 Hz PRF, there is a 17.5% broadening of the 38°C curve compared to the 30°C, with these changes largely occurring from 0 to +6 mm laterally. The 30 Hz PRF curves experienced a greater 27.2% broadening between the 30°C and 38°C conditions, with this broadening also largely occurring from 0 to +6 mm laterally.

Next, differences in cavitation intensity were more closely examined by separately analyzing stable and inertial cavitation values. In Fig. 5, both PRFs show increased SCD and ICD when the baseline tissue temperature increases from 30°C to 38°C. At 15 Hz PRF, SCD and ICD at 38°C increased by 1.8 dB and 4.5 dB, respectively, relative to the SCD and ICD at 30°C (SCD: p < 0.05) (ICD: p < 0.01). Similarly, at 30 Hz PRF, SCD and ICD at 38°C increased by 4.0 dB and 3.5 dB, respectively, relative to the SCD and ICD at 30°C (SCD: p < 0.0001) (ICD: p < 0.001). The raw ICD and SCD values for each temperature-PRF pair are shown in Table I.

Fig. 5.

SCD and ICD at baseline tissue temperatures of 30°C and 38°C at 15 and 30 Hz PRF. SCD and ICD are computed within the ROIs shown in Fig. 4 and are averaged across each 10 s sonication. Significant increases for ICD and SCD at 38°C compared to 30°C occur for both PRFs.

TABLE I.

SCD and ICD Across 30°C and 38°C Baseline Tissue Temperature

	SCD [dB]		ICD [dB]
	30°C	38°C	30°C	38°C
15 Hz PRF	47.4 ± 4.4	49.2 ± 4.2	37.2 ± 7.7	41.7 ± 7.1
30 Hz PRF	47.9 ± 4.2	51.9 ± 4.4	36.6 ± 6.2	40.1 ± 6.4

C. Average Cavitation Activity Does Not Increase at a Constant Baseline Tissue Temperature

To compare changes in cavitation activity across PRF and sonication order, a total average PCI across forty-four (n = 44) sonications per average PCI was calculated for each parameter combination. Fig. 6 (a) and (b) show the averaged PCI in both hindlimbs for 15 Hz and 30 Hz PRF at a baseline tissue temperature of 30°C. At 15 Hz PRF, the maximum intensity of the averaged PCI differs by less than 1 dB between the first and second legs sonicated and the two intensity profiles greatly overlap from −6 mm to 3 mm laterally. At 30 Hz PRF, the averaged first leg PCI has a consistently greater intensity than the second leg. To quantify these differences in intensity within each PCI, the average cavitation intensity for each individual PCI was calculated within the ROI shown. At 15 Hz PRF, the cavitation intensity within the ROI decreased on average by 0.2 dB from the first leg to the second, but this difference was not significant (49.1 ± 2.1 dB at the first leg vs. 48.9 ± 3.1 dB at the second leg; p > 0.05). At 30 Hz PRF, the cavitation intensity within the ROI decreased on average by 2.3 dB from the first leg to the second sonicated, but this difference was also not found to be significant (52.5 ± 8.4 dB at the first leg vs. 50.2 ± 3.0 dB at the second leg; p > 0.05).

Fig. 6.

Averaged PCI at a baseline tissue temperature of 30°C across hindlimb sonication order at (a) 15 Hz PRF and (b) 30 Hz PRF. The location of the hindlimb and the sciatic nerve are indicated in each PCI. The dotted black box represents the ROI from 28 to 32 mm axially and with a 1.7 mm width. There is no great intensity difference between PCI at 15 Hz PRF. At 30 Hz PRF, the first leg sonicated produced on average a more intense PCI, although this intensity increase within the ROI was not significant. The axial intensity profiles illustrate the PCI intensity at 30 mm axially for each PCI at (c) 15 Hz PRF and (d) 30 Hz PRF with the standard deviation shown.

As with the average PCI for temperature and PRF, lateral intensity profiles at the FUS focus of 30 mm axially demonstrate changes in the physical shape of the PCI across sonication order (Fig. 6) (c) and (d). In each intensity profile, the change in cavitation intensity curves across sonication order was determined by computing the FWHM for each curve and then calculating the percent change between the FWHM values. At 15 Hz PRF, there is a 3.0% broadening of the second leg curve compared to the first leg curve. The 30 Hz PRF curves experienced a greater 26.2% broadening between the first and second legs, although unlike the 15 Hz curves the first leg curve is broader than that of the second leg.

Differences in cavitation intensity across sonication order were examined by separately analyzing stable and inertial cavitation values. As with the PRF and temperature data, the averaged SCD and ICD for each sonication order-PRF combination were calculated within the ROI depicted in Fig. 6 (a) and (b). As can be seen in Fig. 7, there are no significant changes in cavitation dose between the first and second limbs sonicated at 30°C. At 15 Hz PRF, SCD decreased by 1.3 dB and ICD increased by 1.3 dB at the second leg sonicated relative to the SCD and ICD at the first leg sonicated, although neither change is statistically significant (SCD: p > 0.05) (ICD: p > 0.05). Similarly, at 30 Hz PRF, SCD and ICD at the second leg sonicated decreased by 1.9 dB and 0.5 dB, respectively, relative to the SCD and ICD at the first leg sonicated, although neither change is statistically significant (SCD: p > 0.05) (ICD: p > 0.05). The raw ICD and SCD values for each sonicated order-PRF pair are shown in Table II.

Fig. 7.

SCD and ICD at a baseline tissue temperature of 30°C at 15 and 30 Hz PRF for both the first and second hindlimb sonicated in each mouse. SCD and ICD are computed within the ROIs shown in Fig. 6 and are averaged across each 10 s sonication.

TABLE II.

SCD and ICD Across Sonication Order at a Baseline Tissue Temperature of 30°C

	SCD [dB]		ICD [dB]
	30°C-1st Leg	30°C-2nd Leg	30°C-1st Leg	30°C-2nd Leg
15 Hz PRF	48.7 ± 3.8	47.4 ± 2.8	36.6 ± 3.0	37.9 ± 2.9
30 Hz PRF	47.9 ± 8.1	46.0 ± 3.4	38.2 ± 4.5	37.7 ± 2.4

D. Average Interframe Displacement Changes with Baseline Tissue Temperature and PRF

To compare the average displacement for each temperature-PRF parameter combination, the three interframe displacement maps taken for each 10 s sonication were analyzed. These displacement values for 15 Hz and 30 Hz PRF at both 30°C and 38°C can be seen in Fig. 8. At 15 Hz PRF, there is a significant increase in average displacement of 1.8 μm between a baseline tissue temperature of 30°C and 38°C (1.7 ± 0.9 μm at 30°C vs. 3.6 ± 3.5 μm at 38°C; p < 0.0001). At 30 Hz PRF, there is a significant increase in average displacement of 1.3 μm between a baseline tissue temperature of 30°C and 38°C (1.7 ± 1.2 μm at 30°C vs. 3.0 ± 1.8 μm at 38°C; p < 0.0001). No significant differences were found between 15 Hz and 30 Hz displacement at 30°C or 15 Hz and 30 Hz displacement at 38°C (p > 0.05).

Fig. 8.

Average interframe displacement during FUS on-time for 15 Hz and 30 Hz PRF at baseline tissue temperatures of 30°C and 38°C.

E. Interframe Displacement Noise Correlates with Inertial Cavitation Dose

Some displacement versus time traces analyzed in the previous section were found to contain very noisy displacement data during the FUS on-time. Specifically, this noise was found to increase with the ICD (Fig. 9) (a). Comparison between the cavitation spectrograms and displacement traces indicated a relationship between the occurrence of broadband cavitation activity and displacement noise. Thus, for each parameter combination in the previous section (n = 1,188 displacement estimations), the displacement noise was plotted as a function of the ICD (Fig. 9) (b). A Pearson correlation coefficient of r = 0.53 was found between the ICD and displacement CV (p < 0.0001), indicating a strong positive correlation between the two parameters. An exponential model of the following form fit the data with an R² value of 0.93:

$$\text{CV}=0.20\cdot \text{exp}(0.39\cdot \text{ICD}[\text{dB}]-17.79)+0.45$$ (2)

Fig. 9.

(a) Example average displacement traces within an ROI around the sciatic nerve. The blue trace illustrates the estimated displacement at a low ICD of 33.3 dB whereas the orange trace illustrates the estimated displacement at a higher ICD of 46.5 dB. (b) Displacement noise, measured by the displacement CV, versus time-averaged ICD for each temperature-PRF parameter combination. An exponential fit, shown by the red dotted line, was plotted with the raw data.

F. Histological Safety Analysis

Of the parameters tested, no indications of damage were found in the negative control or 15 Hz, 30°C experimental condition. Consistent damage was found in the 7.1 MPa positive control condition in the form of tissue swelling, large areas of blood cell extravasation, and clusters of abnormal nerve fibers. A few extravasated blood cells could be seen in the 15 Hz, 38°C and the 30 Hz, 30°C samples, but only in small quantities and not consistently across slides, in contrast to the large areas of extravasated red blood cells that appeared across all slides within the FUS focus in the positive control. Extravasated blood cells as well as a single, small area of abnormal nerve fibers were found at 30 Hz, 38°C, but similarly did not consistently appear across all slides within the FUS focus. Figure 10 presents an image of each condition where the maximum abnormalities were found.

Fig. 10.

Histological H&E staining of the negative control, positive control, 15 Hz PRF experimental, and 30 Hz PRF experimental conditions. Large areas of blood cell extravasation are indicated with black circles. Small clusters of extravasated blood cells are indicated with a black arrow whereas regions of abnormal nerve fibers are indicated with a blue arrow.

IV. Discussion

In this study, we demonstrated the effect of baseline tissue temperature on cavitation activity and nerve displacement during peripheral nerve sonication. Overall, increases in baseline tissue temperature resulted in greater levels of both cavitation and nerve displacement. PCI showed an increase in cavitation intensity within an ROI at the focus between tissue temperatures of 30°C and 38°C. Within the same ROI, both SCD and ICD averaged across the entire sonication period also increased at tissue temperatures of 38°C as compared to levels of SCD and ICD at a temperature of 30°C. In contrast, when both hindlimbs were kept at a baseline temperature of 30°C, no significant differences were found between the PCI cavitation intensity, SCD, or ICD in either limb, regardless of whether the limb was sonicated first or second. These results confirm that changes in cavitation from the first study resulted from changes in tissue temperature and not because the hindlimb sonicated at 38°C was always sonicated second. Taken together, these results show a strong correlation between tissue temperature and the level of cavitation induced via FUS around the sciatic nerve. Along with cavitation activity, nerve displacement also varied with tissue temperature in the same ROI surrounding the sciatic nerve. At both 15 Hz and 30 Hz PRF, greater nerve displacement occurred at a tissue temperature of 38°C compared to the displacement at 30°C. It is important to note that no differences were found between 15 Hz and 30 Hz displacement at either temperature. This is to be expected, as the displacement reported here is measured within a single 5 ms pulse. The nerve will undergo more cycles of displacement across the 10 s sonication at a higher PRF, but the displacement within each 5 ms pulse should not greatly change between PRFs.

The effect of tissue temperature on cavitation and displacement may be the result of changes in tissue viscoelastic properties, such as stiffness and viscosity, as the tissue temperature is modulated. Tissue shear modulus has been shown to decrease as tissue becomes warmer, as long as temperatures are maintained below the point of protein denaturation [33]. Measures of viscosity of soft tissue during heating are scarce, but the viscosity of water decreases with increased temperature [39] and may reflect similar changes in tissue viscosity due to the concentration of water within soft tissues. As tissue softens at higher temperatures, this softer tissue may be more easily displaced by the same ultrasound pressure as compared to stiffer tissue, leading to the increased displacements found at 38°C at both 15 and 30 Hz PRF compared to the displacement found at 30°C. Similarly, as shear stress decreases, this softer tissue may be easier to pull apart compared to stiffer tissue, resulting in lowered thresholds of inertial cavitation [31]. Thus, the increased inertial cavitation activity at 38°C compared to 30°C observed here for both 15 and 30 Hz PRF may be the result of lowered inertial thresholds from tissue softening. Increases in temperature have also been shown to amplify non-linear propagation harmonics in both ex vivo tissue and tissue-mimicking phantoms [40], [41]. The greater SCD at 38°C found at both PRFs, as measured by the RMS amplitude of the 1.1 MHz harmonics, could be influenced by these amplified non-linear propagation harmonics.

This study presents for the first-time evidence of in vivo cavitation modulation in response to changing tissue temperature. This has broad implications for any in vivo study where tissue temperature is impacted by the experimental conditions or where cavitation or displacement control is desired. Accounting for tissue temperature may be important in interpreting fluctuations in cavitation and displacement; similarly, purposeful temperature modulation could be used to promote certain experimental outcomes. From a safety perspective, cooling tissue may be beneficial in in vivo experiments. For example, the use of cooled ultrasound gel, a water bath, or even a cold pack applied to the skin may decrease tissue temperature, especially in small animal models, and thus reduce levels of cavitation or displacement induced in the tissue. Conversely, in instances where cavitation is desired, such as during ablative or lithotripsy procedures, warming the tissue may improve experimental outcomes.

From a safety perspective, the histological findings from this study indicate increased damage at 38°C compared to 30°C in both PRFs, which may arise from greater cavitation within the tissue. At 15 Hz PRF, no abnormalities were found in the tissue sample from consecutive sonications at a 30°C baseline tissue temperature, whereas areas of very minor red blood cell extravasation were found from consecutive sonications at 38°C. In the 30 Hz PRF sample, a few stray red blood cells could be seen near the nerve in the 30°C condition, whereas larger clusters of red blood cells were found near the nerve at 38°C along with an area of abnormal nerve fibers. It is important to note, however, that the damage found in the experimental conditions at both tissue temperatures did not appear consistently across slides within the FUS focus and was highly localized to small areas surrounding the sciatic nerve, unlike the positive control.

To modulate cavitation and displacement using temperature at a similar level in human subjects, more drastic heating or cooling is likely needed than in mice. In humans, increases in limb size will result in temperature changes on the skin surface less prominently affecting the temperature at the level of the nerve; however, temperature modulation is still possible. Studies have shown ice packs applied to human legs for 20 minutes decrease intramuscular temperatures by 4-10°C [42]-[44], and that these changes last for several minutes even after removal of the cold source [42]. Cooling from ice packs has even been shown to increase human muscle tissue stiffness [45]. Baseline temperature and temperature rise from FUS, however, will be more difficult to measure in human subjects, as thermocouples cannot be placed in the tissue without invasive procedures; however, the inclusion of thermal strain imaging into neuromodulatory sequences may allow for non-invasive estimation of temperature rises at the nerve [46].

Since modulating tissue temperature can subsequently modulate levels of both cavitation activity and tissue displacement, tissue temperature could potentially also be controlled to affect neuromodulatory outcomes. Previous studies have supported an acoustic radiation force based mechanism for nerve activation, with evidence including the use of high pressure, short pulses for peripheral nerve activation [18], minimum tissue displacements required for peripheral neural activation via FUS [19], calculated acoustic radiation forces that exceed the thresholds of mechanosensitive ion channels [17], and the link between acoustic radiation force to neural excitation in ex vivo retinal cells [23]. The evidence for cavitation as a neuromodulation mechanism is less clear. Compound action potentials have been found to coincide with recorded inertial cavitation in an ex vivo model [14], [16] and an intramembrane cavitation model has been shown to be capable of reproducing some neuromodulation results [27]-[29]. Stable cavitation has also been linked to increases in membrane excitability by inducing changes in membrane conductance [47]. These studies, however, have focused solely on ex vivo specimen and modeling. This study has demonstrated the induction of cavitation during peripheral nerve sonication in an in vivo model, but as measures of neuromodulation were not taken, this cavitation cannot be mapped directly to any neural excitation or inhibition.

It is worth highlighting the relatively high sonication pressure and MI used in this study, and other peripheral nerve sonication studies, compared to ultrasound neuromodulation conducted in the brain. In this study, mice were sonicated at 4 MPa (3.8 MI). Neuromodulation studies inducing measurable neural activity in the peripheral nervous system have employed high sonication pressures ranging from 4.7-28 MPa and MIs from 2.8-14 [16], [18]-[21]. In contrast, neuromodulatory studies in the brain in both human and animal models typically use much lower pressures, frequently below 1.5 MPa, and tend to stay below an MI of 1.9 [1]-[3], [10], [48]. The need for higher exposure levels to induce neural activity in the peripheral nervous system may be the result of differences in neuron structure, density, and function between the two systems.

Cavitation and displacement in this study were imaged during separate FUS pulses. While this allows for a comparison of cavitation and displacement at close timepoints for high PRF procedures, studies employing neuromodulation at lower PRFs may instead benefit from an interleaving scheme that achieves displacement and cavitation imaging within the same FUS pulse. Since the FUS pulses employed here and in other peripheral neuromodulation studies are on the order of several milliseconds [18], [19], [21], interleaving these imaging schemes within a single pulse is feasible. This interleaving scheme will be investigated in future in vivo studies.

Displacement noise and ICD showed strong correlation across all tissue temperatures and PRFs tested. The association of greater displacement noise with a greater ICD may be the result of broadband activity from cavitation interfering with the acquisition of ¬¬RF data during transmit and receive imaging. The violent collapse of gas nuclei during inertial cavitation emits a broadband white noise signal. This broadband signal could then introduce noise into the RF data during displacement imaging, leading to large disjoints in the cross-correlated data during motion tracking. The occurrence of displacement noise only when the FUS is on supports this hypothesis, as inertial cavitation occurs only in response to the 5 ms FUS pulse and was not found immediately before or after any pulses.

Along with a lack of neuromodulation results, another limitation of the current study is the passive cavitation imaging resolution, specifically in the axial direction. This poor axial resolution complicates the localization of cavitation to the sciatic nerve. The most intense cavitation always occurred around the sciatic nerve and the chosen ROI encompassed only this highest intensity portion of the mapped PCI data; however, better localization would ensure processing of cavitation occurring only at the nerve instead of the surrounding tissue. This axial spreading occurs as a result of the chosen cavitation beamforming algorithm. The TEA algorithm was employed due to its fast computation time compared to other cavitation beamforming algorithms and its ability to process cavitation in the frequency-domain [37], allowing for easy separation of broadband and harmonic frequency components. To improve axial resolution without increasing computational time, coherence-factor based cavitation mapping could be implemented. This beamforming algorithm produces axial resolutions comparable to the eigenspace-based robust Capon beamformer (EISRCB) while keeping computational times similar to that of TEA beamforming [49]. This alternative beamforming method will also be explored in future studies.

V. Conclusion

In this study, cavitation and displacement imaging were simultaneously mapped during peripheral nerve sonication. A strong correlation was found between ICD and displacement noise during FUS on-time, indicating the ability of displacement imaging to serve as a method of both tissue motion tracking and inertial cavitation detection in vivo. Higher levels of cavitation activity and nerve displacement were found at normal body tissue temperatures; tissue temperature modulation may therefore be used as a method of selectively promoting or hindering inertial cavitation and tissue displacement in vivo. Future studies will use this method of interleaving to compare both cavitation activity and nerve displacement to sciatic nerve activation during the same FUS neuromodulation experiment, thus determining whether inertial cavitation correlates with in vivo neuromodulation in addition to nerve displacement. Tissue temperature modulation will also be investigated as a method of promoting neuromodulatory success.

Biographies

Erica P. McCune (Graduate Student Member, IEEE) received the B.S. degree from the joint department of biomedical engineering at North Carolina State University, Raleigh, NC, USA and the University of North Carolina at Chapel Hill, Chapel Hill, NC, USA in 2020 and the M.S. degree in biomedical engineering at Columbia University, New York, NY, USA in 2022. She is currently pursuing the Ph.D. in biomedical engineering at Columbia University.

Her research interests include peripheral focused ultrasound neuromodulation for the treatment of neuropathic pain and better understanding the mechanisms of peripheral neuromodulation.

Stephen A. Lee (Member IEEE) received the B.S. degree in biomedical engineering from the University of North Carolina at Chapel Hill, Chapel Hill, NC, USA in 2015, the M.S. degree in biomedical engineering from Columbia University, New York, NY, USA in 2018, and the Ph.D. from Columbia University, New York, NY, USA in 2022. His thesis focused on developing imaging and therapeutic ultrasound neuromodulation of the peripheral nervous system in the context of neuropathic pain.

As of 2022, he is a Vanier-Banting Postdoctoral Fellow with the Department of Engineering Physics, Polytechnique Montreal, Montreal, QC, Canada. His research interests include how neurovascular flow and function can be used as brain biomarkers of cerebral health and serve as early indicators of neurodegeneration.

Dr. Lee was a recipient of the IEEE IUS Student Paper Competition Award in 2019.

Elisa E. Konofagou (Fellow, IEEE) is the Robert and Margaret Hariri Professor of Biomedical Engineering and Professor Radiology as well as Director of the Ultrasound and Elasticity Imaging Laboratory at Columbia University in New York City. Her main interests are in the development of novel elasticity imaging, theranostic and therapeutic ultrasound methods for the advancement of therapeutic ultrasound. Her group has been working in the ultrasound field with applications in cardiovascular elasticity imaging, breast and pancreatic cancer diagnosis, chemotherapy treatment and ablation monitoring as well as brain and nerve therapeutics that involve drug delivery through the blood-brain barrier. Elisa has co-authored over 280 published articles in the aforementioned fields. She is recipient of the CAREER award by the National Science Foundation (NSF), the Nagy award by the National Institutes of Health (NIH), the Technological Achievement Award by the Institute of Electrical and Electronic Engineers (IEEE) Engineering in Medicine and Biology society (EMBS), the Carl Hellmuth Hertz Ultrasonics Award by the IEEE Society in Ultrasonics, Ferroelectrics and Frequency Control (UFFC) gas well as additional recognitions by the American Heart Association, the Acoustical Society of America, the American Institute of Ultrasound in Medicine, the American Association of Physicists in Medicine, the Wallace H. Coulter foundation, the Bodossaki foundation, the Society of Photo-optical Instrumentation Engineers (SPIE) and the Radiological Society of North America (RSNA).

Elisa is an Elected Member of the National Academy of Medicine (US), an Elected Fellow in multiple organizations including the IEEE, the American Institute of Biological and Medical Engineering (AIMBE) and of the Acoustical Society of America (ASA). She is an administrative committee member of the IEEE in Engineering in Medicine and Biology conference (EMBC) and technical committee of IEEE in Ultrasonics, Ferroelectrics and Frequency Control Society (UFFC), the Acoustical Society of America and the American Institute of Ultrasound in Medicine. Prof. Konofagou is also a technical committee member of the Acoustical Society of America, the International Society of Therapeutic Ultrasound, the IEEE Engineering in Medicine and Biology conference (EMBC), the IEEE International Ultrasonics Symposium and the American Association of Physicists in Medicine (AAPM). Elisa serves as Associate Editor in the journals of IEEE-UFFC Transactions, Ultrasonic Imaging and Medical Physics. Elisa also serves as a member of the Greek National Council for Research, Technology and Innovation which directs all scientific research and innovation in Greece and she recently served as President of the Focused Ultrasound Symposium held in Bethesda, Maryland. She is the President-Elect of the International Society of Therapeutic Ultrasound.